- Tuesday March 23 2010, 08:34 -
sjcarbon: sorry--couldn't reconnect with the other one
sjcarbon: i guess no chris though.
- sjc has disconnected (Ping timeout)
sjcarbon: by the way, for people who haven't had contact before, here's Sven (svenmh).
gwg: hi sven
sjcarbon: He's been working on the phylotree stuff for AmiGO.
pasc: nice !
svenmh: I'm at Princeton U, doing what Seth said.
sjcarbon: sven, i'm going to try the loader today--i'll let you know if there are any problems.
Ben: hi sven, I guess we'll meet you next week
- rama has joined the room
rama: sorry i am late
sjcarbon: no problem--i was having connection problems anyways.
- cjm has joined the room
rama: same here-connection issues
sjcarbon: i guess the first item is a question for ben. do you know how amigo is doing with the connection overload?
sjcarbon: is seems like it has been ongoing since late last week?
rama: i don't know really
Ben: not a clue
- 08:39 -
sjcarbon: no then, i'll bother go-admin for it then.
rama: Gail said something about increasing user connetion to 60
Ben: I must admit, I have not looked into it.
sjcarbon: i haven't read email yet, so i guess we'll see later on.
sjcarbon: no problem. hopefully not in the code if it started suddenly. and probably not a ddos.
sjcarbon: next item.
sjcarbon: i've switched in the new GOOSE and there doesn't seem to be any problems.
sjcarbon: if people here use GOOSE, please keep an eye on the links that it generates--that's probably where issues might come up.
rama: I wouldn't know what to look for
rama: Seth- what exactly changed in GOOSE
rama: I use it, but not often though
sjcarbon: i switched the whole thing out for a rewrite using the new code.
rama: I am assuming the numbers retrieved match etc (you had some internal tests to compare?)
sjcarbon: of course.
- 08:44 -
rama: okay. thanks
sjcarbon: but that part is still just passing the sql through.
Ben: does it use DBIx::Class or Moose?
sjcarbon: some part probably uses DBIx::Class deep down, but only as a dependency to the AmiGO framework.
sjcarbon: it'll use more in the future as features move in.
sjcarbon: and features can move in now due to the rewrite.
sjcarbon: anyways, new goose is out.
rama: great. thanks
sjcarbon: moving on, news from the world of the phylotree.
sjcarbon: can i pass this to you sven?
- 08:49 -
svenmh: this is the current working code that is in cvs
svenmh: it might be slow, that system gets used for lots of things.
sjcarbon: about to get slower.
pasc: very nice !
svenmh: but I've got to work matching the ids better.
svenmh: but I've gotten more insight, and the loading script should get better in the next few days.
rama: nice. thanks Sven
pasc: so, the list is organized by date last annotated descending?
rama: I am not getting the big picture though (how things are going to be integrated etc)
rama: I am sure more stuff is coming.
svenmh: yea, currently order by last annotation
rama: Displaying the name of the family next to the PANTER ID will be useful
cjm: seth and i can work with you on the speed issue
svenmh: hopefully most of the panther ids will go away and I get better matches
- 08:54 -
cjm: it looks like the members are panther IDs rather than gene product IDs from the GO DB, is this the loading script issue you mention?
svenmh: it looks like i'm going to have to treat each organism a little differently.
svenmh: Yeast will get matched to gene gene_product row, and human to a protein row. Plan on looking at other organism today.
svenmh: currently where you see pthr ids are the matches I missed.
svenmh: I think that is it from me, unless there are questions.
Ben: This might be a feature for later versions, but I think people would like to see a cross reference with Pfam and other domain-based methods where appropriate.
rama: I don't know the scope of the phylo-tree
pasc: did we decide how to integrate the Date selected for concurrent annotation?
Ben: and since I am throwing out random work for people to do...
rama: But when you are ready i would like to offer some suggestions for displayes
Ben: I think that a phylogeny (kingdom/phyum/genus etc) level filtering system would be really cool.
- 09:00 -
Ben: So you could say "show me all panther families that have vertebrate members"
rama: yes, I have been pushing that for other features in AmiGO
rama: (filtering by kingdom i mean)
- sjc has joined the room
cjm: Ben - your wish is our command.
svenmh: Currently, what I'm showing is what Kara (my boss) wants. After I make her happy I'll add other features.
rama: If there is a wish list that you guys maintain, I would like to add to that
svenmh: I could keep them is pod docs too. I did write pod docs for most of it.
- 09:05 -
Ben: it's not a wish, just a suggestion. I can pretty much guarentee that I will never use the tool for any actual work, so don't change it to please me.
cjm: For now people should just email suggestions to the wpwg.
rama: we shd run it by curators
rama: yes, I agree with Chris
cjm: shall we also cover live search and term lineage
cjm: (20 mins to go, I have to leav a little earlier)
sjcarbon: moving on to live search then.
rama: I see the homolog filter
sjcarbon: presented now as an alternative search in the "more tools" area.
rama: what are the numbers here- Total: 2 First: 1 Last: 2
- 09:10 -
sjcarbon: maks more sense if you have more than 10 results.
rama: I searched for 'topo' using the gene prodcut search and i saw this on the right side
rama: hah. got it
cjm: Seth, I know we keep going round on this, but I still find it unintuitive how if you type "neurog*" that "neurogenesis" is _not_ the top hit
sjcarbon: that is fixed for completion.
sjcarbon: i guess i didn't port that code over to this...
sjcarbon: i'll look into it.
sjcarbon: (no code shared for that part.)
cjm: ok good
rama: (another example- i typed autopha for autophagosome and did not get anything)
sjcarbon: use '*' afterwards
rama: i see
sjcarbon: the completion does that automatically.
rama: you mention that next to the search box
rama: star works. thanks
rama: but you shd mention that
sjcarbon: i'll make a note about that when you return no results.
sjcarbon: 15 minutes to go. anything else here? people can always get me on the wpwg list.
- 09:15 -
Ben: here is another one ... if you type in protein kinase or protein kinase activity
Ben: you don't get the exact match first. YOu get GO:0004713 protein tyrosine kinase activity
Ben: I wonder if it would make sense to but a Levenstein distance sort at the very end.
Ben: sorry typo
sjcarbon: likely not for this version.
Ben: again, I am just speaking theoretically
sjcarbon: i'll retune it to be more like the completion.
sjcarbon: also, quotes help.
gwg: "the brightest flame burns quickest"?
sjcarbon: i feel much better now.
gwg: carpe diem cras enim moriemur!
sjcarbon: enough of the advanced features come out of the box in later versions of lucene, it is probably not worth it to try to backport the stuff.
- 09:21 -
sjcarbon: some cleverness. can be done with the search syntax as well:
sjcarbon: anything else here?
sjcarbon: onto the term lineage view.
sjcarbon: i haven't touched this since last week.
sjcarbon: waiting for all of the comments to come in.
sjcarbon: i think people here are generally well-disposed towards it probably, so maybe there isn't much to talk about here in chat.
sjcarbon: any dislikers here?
rama: may be you shd post it on the GO mailing list like the tree view
pasc: good idea
rama: the live search
sjcarbon: whoops. wrong link--i was trying this:
- 09:27 -
sjcarbon: now my comments should make more sense.
rama: the live search is cool. please post a message on the gomailing list for more comments
sjcarbon: will do after some fixes.
sjcarbon: almost out of time.
sjcarbon: i don't know if we'll be able to resolve the term lineage bofore the meeting.
- pasc has disconnected (Read error: EOF from client)
- cjm has disconnected (Quit)
sjcarbon: it that something we should hash out here, or add it to the meeting agenda?
- gwg has disconnected (Quit: Leaving.)
rama: meeting ageda
sjcarbon: leadership fail.
sjcarbon: okay. meeting agenda. i hope i can be more convincing if i can wave my arms around.
sjcarbon: thanks, bye!
rama: good luck
sjcarbon: rama--next monday?
rama: monday? what?
sjcarbon: i thought you wanted to meet before the meeting.
rama: meeting starts Tue
rama: I don't know if you want to drive down
rama: on monday
- 09:32 -
sjcarbon: i have no problem if there is stuff you want to go over.
sjcarbon: let's see how this week goes then.
rama: nothing in particular
sjcarbon: okay, see y'all soon then
rama: btw, we are in 1501 Cal ave
rama: not on cmapus any more
sjcarbon: better or worse parking?
rama: awesome parking
- rama has left the room
- Ben has disconnected (Quit: Quitting!)
sjcarbon: see ya!